CC2 Lambda equations

[obackhouse]

The T2 amplitudes in CC2 require

< e(-T2) f e(T2) > + < e(-T1) v e(T1) >

So I assume the L2 amplitudes are just

< (1 + L1 + L2) e(-T2) f e(T2) > + < (1 + L1 + L2) e(-T1) v e(T1) >

But, then when I calculate DMs as

< (1 + L1 + L2)  e(-(T1 + T2)) e(T1 + T2) >

I don't recover the same electronic energy as I get when I just optimise the T amplitudes (which of course works fine for CCSD). Can't work out what I am missing and can't find much literature on Lambda-CC2. Not sure whether to consider lambdas and DMs in CC2 bugged, or whether this is just a deficiency in the theory.

[obackhouse]

Just to make note of things I've tried, after discussing with Charlie he pointed out that formally in CC2 derivations there's a f-T1 term in the doubles amplitude too - this is unconnected though so contributes nothing, and that's also true for the lambda equations.